Non-Von
Non-Von is a Dartmouth-affiliated semiconductor startup building neuromorphic AI accelerator chips that eliminate the von Neumann memory bottleneck via per-core dedicated memory, targeting defense, data center, and autonomous systems customers with up to 100x power efficiency versus GPU systems.
- Company typePrivate
- Founded-
- HeadquartersNew Hampshire, United States
- Headcount1–10
- GTM typeB2B
- OfferingHardware or Manufacturing
What Non-Von does
Non-Von, LLC is a Dartmouth-affiliated semiconductor startup developing a neuromorphic AI accelerator architecture that eliminates the traditional von Neumann memory bottleneck by pairing each processing core with its own dedicated memory. The company's core technology centers on the "novo" architecture, with a proof-of-concept chip (novo1) fabricated in 2024 on a 28nm TSMC process and demonstrated in a fielded commercial edge-AI product. A substantially more powerful successor (novo2) is planned for the end of 2026. The company complements its hardware with the SCULPT software toolset, which automatically converts trained models from PyTorch, ONNX, Keras, TensorFlow, SKLearn, and NumPy into sparse, compacted implementations optimized for Non-Von silicon.
The company targets enterprise and government customers in three primary verticals: defense and military (low-power AI for autonomous drones and tactical edge devices), data centers (energy and cooling cost reduction versus GPU-based systems), and autonomous systems (real-time edge inference for robotics, land vehicles, and aerial drones). Go-to-market is direct enterprise engagement, with the company offering custom silicon fabrication, mass-produced programmable AI accelerator boards (M2 form factor), FPGA development boards, and SDK/software licensing. Pricing is quote-based and not publicly disclosed.
Non-Von is a privately held LLC headquartered in Hanover, New Hampshire, operating as an all-US remote team of approximately 9 employees. Leadership includes founder and President Richard Granger (Dartmouth Professor and Director of the Dartmouth Brain Engineering Lab), founder and CTO Eli Bowen, CEO Derek Odom (retired US Army Special Forces Lieutenant Colonel), and Chief Counsel Rick Brock. The company has not disclosed any funding rounds, institutional investors, or named customers, and is currently not actively hiring.
Non-Von firmographics
Firmographics- Name
- Non-Von
- Legal name
- NON-VON, LLC
- Website
- https://non-von.com
- Company type
- Private
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- Non-Von is a Dartmouth-affiliated semiconductor startup building neuromorphic AI accelerator chips that eliminate the von Neumann memory bottleneck via per-core dedicated memory, targeting defense, data center, and autonomous systems customers with up to 100x power efficiency versus GPU systems.
- Ownership category
- akta.pro rank
Non-Von industry classification
Industry- Product category
- Neuromorphic AI Accelerator Hardware
- NAICS
- Electronic Computer Manufacturing (334111)
- SIC
- Semiconductors & Related Devices (3674)
- akta.pro primary industry
- AI Accelerators (GPUs/TPUs/NPUs/ASICs) (HDAAAAAA)
- akta.pro secondary industries
- AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers) (HDAAAAAI), AI/Neural Processing Units (NPUs) & ML Accelerators (HDAHAJAF), Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC) (HDAHABAC), Edge AI Hardware & NPUs (SoCs, modules, accelerators) (HDAAAJAC)
Keywords
Where Non-Von is headquartered
LocationHeadquarters
- HQ city
- New Hampshire
- HQ country
- United States
- HQ region
- North America
Offices1 record
Markets served
Non-Von business model
Business model- GTM type
- B2B
- Offering type
- Hardware or Manufacturing
- Cost components
- Technology or R&D, Supply Chain, Personnel, Operations, Marketing or Sales
Revenue model
- Custom Silicon Sales: Non-Von fabricates custom silicon chips for customers. Models can be printed as custom Non-Von silicon tailored to specific applications.
- Programmable AI Accelerator Boards: Mass-produced programmable AI accelerator boards (M2 form factor) that can be imprinted with customer models. These include M2 FPGA boards and USB development adapters.
- SDK/Software Licensing: Non-Von's SCULPT software toolset enables model optimization on their hardware. Revenue may include SDK licensing for their development tools and compiler technology.
Go-to-market motion2 records
Distribution channels2 records
Marketing channels3 records
Non-Von product offering
Product offeringCore offering
Non-Von designs, fabricates, and sells neuromorphic AI accelerator chips and accompanying software tools. Its novo architecture removes the von Neumann memory bottleneck by pairing each processing core with its own dedicated memory, achieving up to 100x power efficiency on AI inference versus GPU-based systems. Offerings include the novo*1 proof-of-concept chip, FPGA development boards in M.2 form factor, mass-produced programmable AI accelerator boards, custom silicon fabrication services, and the SCULPT SDK that ports models from PyTorch, ONNX, TensorFlow, Keras, SKLearn, and NumPy.
Product overview
Non-Von is a Dartmouth-affiliated startup offering a neuromorphic AI hardware-software platform centered on its novel novo architecture. The platform combines the novo*1 and forthcoming novo*2 chips with the SCULPT software toolset to deliver AI inference acceleration with up to 100x power efficiency compared to GPU-based systems. The architecture entirely eliminates the traditional von Neumann memory bottleneck by pairing each processing core with dedicated memory. The product line includes proof-of-concept silicon (novo*1), planned next-generation chips (novo*2), FPGA development boards in M.2 form factor, and programmable AI accelerator boards. The SCULPT toolset automatically converts models from common frameworks (PyTorch, ONNX, Keras, TensorFlow, SKLearn, NumPy) into sparse, compacted implementations optimized for Non-Von hardware.
Differentiator
Problem solved
Functional benefit
Products and services
- novo*1 Chip Non-Von's first neuromorphic AI accelerator chip, fabricated at 28nm via TSMC. Eliminates the von Neumann bottleneck by pairing each processing core with its own dedicated memory, achieving 100x power efficiency on sparse AI models versus GPU-based systems, and is targeted at edge AI inference and data center deployments.
- FPGA V2 Development Board An FPGA-based development board packaged in M.2 form factor with USB adapter that enables developers to prototype, test, and validate novo architecture designs and AI model implementations before production silicon fabrication.
- M.2 FPGA Development Board An M.2 form factor FPGA board for Non-Von chip development and testing, designed for seamless integration into existing systems and edge AI prototyping.
- Programmable AI Accelerator Boards Mass-produced programmable AI accelerator boards in M.2 form factor into which trained models can be imprinted, offering customers a flexible alternative to custom silicon for deploying Non-Von-optimized AI models.
- Custom Silicon Fabrication A custom silicon fabrication service in which Non-Von prints customer-trained AI models as custom Non-Von ASIC silicon tailored to specific applications and deployed in volume.
- SCULPT (Sparse Compact Ultra-Low Power Toolset) Software toolset that automatically optimizes trained AI models for novo hardware by applying sparse compaction techniques; supports PyTorch, ONNX, Keras, TensorFlow, SKLearn, NumPy and other frameworks with just a few lines of code, enabling implementation on Non-Von silicon or programmable boards.
Quantifiable outcome
- 100x power efficiency improvement on tasks from deep-learning visual recognition to transformer-based chatbots
- +4 more outcomes
Companies that use Non-Von
Customer profileNamed customers3 records
Segments4 records
Ideal customer profiles3 records
Non-Von technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
Integration6 records
AI capability7 records
Feature5 records
Non-Von partnerships and signals
Strategic signalScale indicators5 records
Recent moves5 records
Expansion highlights6 records
Non-Von competitors and assessment
Company assessmentDirect peers
- Syntiant: Low-power edge AI processor company for always-on devices. Comparable to Non-Von in the ultra-low-power edge AI inference market, with similar competition against Google Edge TPU and Coral.
- Hailo: Edge AI processor company offering high-performance, low-power inference chips. Direct peer in the edge AI accelerator market, with similar focus on power efficiency for vision and edge workloads.
- Graphcore: AI accelerator (IPU) startup pursuing a novel architecture for AI workloads. Comparable to Non-Von as a fabless chip company developing non-GPU architectures targeting efficiency gains for AI inference.
- Groq: AI accelerator chip company focused on low-latency inference. Direct competitor to Non-Von in the AI accelerator silicon market, targeting similar efficiency-per-watt advantages over GPUs.
- SambaNova Systems: AI accelerator chip company offering enterprise inference and training hardware. Directly comparable as a deep-tech AI chip startup competing in the custom silicon market against NVIDIA.
- Cerebras Systems: Fabless AI accelerator company with novel wafer-scale architecture. Comparable to Non-Von as a fabless AI chip startup pursuing architectural alternatives to GPUs for AI compute.
- Mythic AI: Analog compute-in-memory AI chip startup focused on edge inference. Comparable to Non-Von in targeting edge AI with novel memory architectures to reduce power consumption.
- BrainChip: Neuromorphic processor company for edge AI. Closely comparable to Non-Von's neuromorphic, brain-inspired architecture paradigm and edge AI positioning.
Broad incumbents
- NVIDIA: Dominant incumbent in GPU-based AI compute with the Jetson edge lineup and data center GPUs. The primary competitor Non-Von benchmarks against and the established ecosystem Non-Von is challenging.
- Google (Edge TPU / Coral): Google's Edge TPU is a direct competitor in the low-power edge AI inference market that Non-Von benchmarks against (claiming 400-800% efficiency advantage over Coral Dev Board).
Market position
Strengths4 records
Weaknesses4 records
Competitive moat3 records
Key risks7 records
Key highlights7 records
Customer concentration
Non-Von social profiles
Digital presenceNon-Von financial estimates
Financial estimateRevenue estimate
Valuation estimate
Non-Von leadership team
Management profileNumber of profiles
Profiles4 records
Non-Von funding detail
Funding detailFunding overview
Funding rounds
Investors
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Non-Von M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Non-Von
What does Non-Von do?
Non-Von designs, fabricates, and sells neuromorphic AI accelerator chips and accompanying software tools. Its novo architecture removes the von Neumann memory bottleneck by pairing each processing core with its own dedicated memory, achieving up to 100x power efficiency on AI inference versus GPU-based systems. Offerings include the novo*1 proof-of-concept chip, FPGA development boards in M.2 form factor, mass-produced programmable AI accelerator boards, custom silicon fabrication services, and the SCULPT SDK that ports models from PyTorch, ONNX, TensorFlow, Keras, SKLearn, and NumPy.
Is Non-Von a public or private company?
Non-Von is a private company. It is classified as founder individual operated bootstrapped and is currently operating.
When was Non-Von founded?
Non-Von was founded in -1. It employs 1 to 10 people.
Where is Non-Von based?
Non-Von is headquartered in New Hampshire, United States, in the North America region.
How does Non-Von make money?
Three revenue lines are on record. Custom Silicon Sales are the primary driver. The others are programmable AI Accelerator Boards and SDK/Software Licensing.
Who are Non-Von's main competitors?
Direct peers on record are Syntiant, Hailo, Graphcore, Groq, SambaNova Systems, Cerebras Systems, Mythic AI and BrainChip. Broad incumbents are NVIDIA and Google (Edge TPU / Coral).
Does Non-Von have an API?
No public API is recorded for Non-Von.
What industry is Non-Von in?
Non-Von's product category is Neuromorphic AI Accelerator Hardware. Its primary akta.pro industry code is HDAAAAAA, AI Accelerators (GPUs/TPUs/NPUs/ASICs), with a secondary code of HDAAAAAI, AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers). Its NAICS code is 334111 and its SIC code is 3674.